Running prediction on webcam. Invalid Darknet command: detector

[NatalliSu]

Hello. I have installed Darknet on Windows 10 step by step as in the description. When I run it on my webcam I got an error message invalid Darknet command: detector. The command line that I used and output is bellow.

darknet detector demo test.data yolov4-tiny.cfg yolov4-tiny.weights -c 0

←[1;33mskipped validating of argument #1 "demo" (does not appear to be a valid parameter, file, or directory)←[0m
Darknet V3 "Jazz" v3.0-193-g0a3dca26
CUDA runtime version 12060 (v12.6), driver version 12070 (v12.7)
cuDNN version 12060 (v9.5.1), use of half-size floats is ENABLED
=> 0: NVIDIA GeForce RTX 3080 Laptop GPU [#8.6], 16.0 GiB
OpenCV v4.9.0, Windows 10 Pro * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

• Error location: C:\src\darknet\src-lib\detector.cpp, run_detector(), line #2196
• Error message: invalid Darknet command: detector
• Thread #20748: main darknet thread
• Version v3.0-193-g0a3dca26 built on Nov 12 2024 10:23:45

---

backtrace (7 entries):
1/7: $☻()
2/7: $☻()
3/7: $☻()
4/7: $☻()
5/7: $☻()
6/7: BaseThreadInitThunk()
7/7: RtlUserThreadStart()
known threads: 1
1/1: #20748: main darknet thread

If I run detector test everything is fine, but detector demo causing error message above.

[stephanecharette]

darknet detector demo test.data yolov4-tiny.cfg yolov4-tiny.weights -c 0

That is an invalid command. Where did you get that command from?

The new V3 commands are listed in the darknet --help output and in the readme: https://github.com/hank-ai/darknet?tab=readme-ov-file#cli

So the command you probably want would be something near the following (adjust as necessary):

darknet_08_display_webcam yolov4-tiny.cfg coco.names yolov4-tiny.weights

[NatalliSu]

The command
darknet_08_display_webcam yolov4-tiny.cfg coco.names yolov4-tiny.weights
works perfectly. Thank you.

The command I was talking about (detector demo) I copied from the readme link you mentioned and adjusted file names. Now I see, it was an old version. Probably, that is why I got the error message.